pygformula 1.0.5

A python implementation of the parametric g-formula

pygformula: a python implementation of the parametric g-formula

Overview

The pygformula package implements the non-iterative conditional expectation (NICE) algorithm of the g-formula algorithm (Robins, 1986). The g-formula can estimate the counterfactual mean or risk of an outcome under hypothetical treatment strategies (interventions) when there is sufficient information on time-varying treatments and confounders.

Features

• Treatments: discrete or continuous time-varying treatments.
• Outcomes: failure time outcomes or continuous/binary end of follow-up outcomes.
• Interventions: interventions on a single treatment or joint interventions on multiple treatments.
• Random measurement/visit process.
• Incorporation of a priori knowledge of the data structure.
• Censoring events.
• Competing events.

Requirements

The package requires python 3.8+ and these necessary dependencies:

• cmprsk
• joblib
• lifelines
• matplotlib
• numpy
• pandas
• prettytable
• pytruncreg
• scipy
• seaborn
• statsmodels
• tqdm

Documentation

The online documentation is available at pygformula documentation.